Capital Allocators – Inside the Institutional Investment Industry Daniel Mahr – Glass Box Quant at MDT Advisers (EP.472)
15 snips
Nov 20, 2025 
Daniel Mahr, head of the quantitative equity investing group at Federated Hermes, shares insights from his extensive career. He discusses his journey from flipping IPOs to developing sophisticated decision tree models for stock selection. Mahr explains the balance between analytical rigor and human judgment, emphasizing a 'glass box' approach to transparency. He also explores challenges like overfitting and the impact of market changes on strategies, highlighting how quirky factors influence outcomes. Mahr's passion for evolving models fuels continuous improvement in the unpredictable world of investing.
AI Snips
Chapters
Books
Transcript
Episode notes
Decision Trees Tailor Questions By Company Type
- Decision trees ask sequential yes/no questions to tailor forecasts to subgroups of companies.
- This lets the model use different questions for companies with different characteristics for more precise predictions.
Glass Box Over Black Box
- MDT favors transparency and calls their approach a 'glass box' rather than a black box.
- They prioritize being able to see and explain what drives model decisions daily.
Avoid Both Overfitting And Underfitting
- Balance model complexity to avoid overfitting and underfitting; both are harmful.
- Use techniques and judgment to strike the right model complexity for noisy market data.
